↔ [Converter] Add support for aten::__derive_index in Torch-TensorRT

[p1x31]

• Function Schema:
  aten::__derive_index(int index, int start, int step) -> (int)

• Original PyTorch API:

• https://github.com/pytorch/pytorch/blob/master/torch/csrc/jit/runtime/register_prim_ops.cpp#L183

• Relevant TensorRT Documentation:

Alternatives

Additional context

pytorch/pytorch#48019

[narendasan]

What is the actual functionality here? Is it like start + index * step?

[p1x31]

Yes. An error only occurs when using Torch-TensorRT in Python like this. When compiling with torch.jit.script separately and passing TorchScript module to Torch-TensorRT everything works fine.

[Pikassik]

also faced this problem

[github-actions]

This issue has not seen activity for 90 days, Remove stale label or comment or this will be closed in 10 days